Ok, I'm running Fedora Core 2 on a P-II 333 MHz box w/ 256 MB of RAM and a 20GB HDD. The box is serving as a proxy server for our DSL Internet connection and serves about 15 or so machines. The CPU load averages are 0 and no swap is being used even though 2GB of swap is reserved. Squid 2.5-STABLE7 is being used as the caching HTTP proxy. Cool.
Here's the situation: normally when accessing websites through the proxy, sites load fast. If content is found in the Squid cache, it loads even faster. Even with 10 users connected to the proxy, performance is excellent.... UNTIL for some reason proxy connections seem to "stall" and the browser will appear to "spin" for a bit. Then suddenly the problem resolves itself and all is well again. Strange, indeed.
I'm trying to troubleshoot this without much success. Sometimes, the lag is due to slow DNS lookups or it at least appears to sometimes be the case. Sometimes, the lag is due to the remote site being busy. I confirm this by access the Squid cachemgr to get info on the state of Squid and that responds very quickly, which leads me to believe the remote site is sometimes the culprit. The most bizarre problems involve "stalled" access to the Squid cachemgr as well. IN some cases, I'll try to access a fast site, like Amazon.com or something like that,and the browser will just "spin". Then I'll try to access the Squid cachemgr and that access will "spin" too! Then everything will load quickly.
When the browser "spins" like this, I ping various sites from the proxy itself (logged in via SSH) to see if I can isolate the bottleneck and sometimes the pings won't return any replies! I can ping a raw IP address and get no replies and suddenly get replies back starting with seq #40 or something, meaning the first 39 packets didn't get successfully sent.
I've got LinkSys (I believe) PCI adaptor (eth0) (uses the Tulip driver) and a 3COM 3C905B PCI adaptor (eth1) in the box. The LinkSys adaptor is operating in half-duplex mode and is connected to the LAN and the 3COM card is connected to the Netopia DSL modem.
Here is ifconfig output:
Code:
eth0 Link encap:Ethernet HWaddr 00:04:5A:64:93:28
inet addr:192.168.5.2 Bcast:192.168.5.255 Mask:255.255.255.0
inet6 addr: fe80::204:5aff:fe64:****/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:159953 errors:0 dropped:0 overruns:0 frame:0
TX packets:249455 errors:3 dropped:0 overruns:0 carrier:3
collisions:11125 txqueuelen:1000
RX bytes:28852003 (27.5 Mb) TX bytes:280877798 (267.8 Mb)
Interrupt:10 Base address:0xd000
eth1 Link encap:Ethernet HWaddr 00:10:5A:65:67:EA
inet addr:65.104.xx.yyy Bcast:65.104.xx.255 Mask:255.255.255.0
inet6 addr: fe80::210:5aff:fe65:****/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:240640 errors:0 dropped:0 overruns:0 frame:0
TX packets:208118 errors:0 dropped:0 overruns:0 carrier:2
collisions:68 txqueuelen:1000
RX bytes:276291161 (263.4 Mb) TX bytes:32145441 (30.6 Mb)
Interrupt:5 Base address:0xd000
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:8062 errors:0 dropped:0 overruns:0 frame:0
TX packets:8062 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:1521910 (1.4 Mb) TX bytes:1521910 (1.4 Mb)
Any ideas on how to best troubleshoot this?
Thanks....
Peace...